An atomic broadcast or “total order broadcast” may be used in fault tolerant distributed systems to ensure that multiple distributed processes receive operations in an equivalent sequence, regardless of which node in the distributed system initiates each of the operations. The operations may be propagated to each node in the distributed system such that either each operation completes at each node, or the operation is rolled back at each node.